Surely that has been addressed before. Is there any reason that a blind test has to be short term?
It is related to short and long term auditory memory. The length of the test depends on what you are looking for, and surely the practical limitations. The best duration can even be determined experimentally - you carry tests to show the difference you are looking for using cases with positive differences and null differences with different time lengths. Then you analyze the data and choose as best the one that gives you more accurate results - a balance of more positive identifications and less spread around the results. Happily you do not have to do it every time - it was studied and analyzed by many people, you just have to dig in their papers.
Just one note - do not try to get in blind testing if you are not prepared to carry proper statistical analysis of it, and understand statistical requirements and limitations. Remember that only a few became millionaires in Massachusetts
http://www.boston.com/news/local/massachusetts/articles/2011/07/31/a_lottery_game_with_a_windfall_for_a_knowing_few/?page=full